Characterization of metabolism and in vitro permeability study of notoginsenoside R1 from radix notoginseng

Article abstract of DOI:10.1021/jf1005885

As a main and characteristic constituent in Radix notoginseng, the fate of notoginsenoside R1 (NGR1) in human is largely unknown. The present study investigated, for the first time, NGR1 metabolism by human intestinal bacteria and liver subcellular fractions, and permeability properties of NGR1 and resultant metabolites on a Caco-2 model. Samples were qualitatively analyzed using HPLC-MS/MS and quantitatively determined using HPLC-UV. When incubated with pooled human intestinal bacteria anaerobically, NGR1 showed biphasic elimination: an insignificant decrease in the first 8 h followed by a rapid elimination during 8-48 h. Four metabolites, three unambiguously identified as ginsenosides Rg1, F1 and 20(S)-protopanaxatriol formed via stepwise deglycosylation, and one tentatively assigned as a dehydrogenated protopanaxatriol with transformation occurring at the tetracyclic triterpenoid skeleton, were produced sequentially. Rg1 and F1 were formed transiently at low apparent velocities, while 20(S)-protopanaxatriol was the major metabolite with a formation rate close to the rate of NGR1 elimination and a low elimination rate. NGR1 remained intact in human liver S9 or microsomes over 1 h. Transport study of NGR1 and its metabolites revealed an ascending permeability order with stepwise deglycosylation. Taken together, the results revealed a determinant role of intestinal bacteria in the overall disposition and potential bioactivity of NGR1 in human.

Full text of DOI:10.1021/jf1005885

5770 J. Agric. Food Chem. 2010, 58, 5770–5776
DOI:10.1021/jf1005885
Characterization of Metabolism and in Vitro Permeability Study
of Notoginsenoside R1 from Radix Notoginseng
J
IAN-QING
R
UAN, WENG-I
M
L
EONG, R
U
Y
AN,* AND
Y
I
-TAO
WANG
Institute of Chinese Medical Sciences, University of Macau, Taipa, Macau SAR, China
As a main and characteristic constituent in Radix notoginseng, the fate of notoginsenoside R1
(NGR1) in human is largely unknown. The present study investigated, for the first time, NGR1
metabolism by human intestinal bacteria and liver subcellular fractions, and permeability properties
of NGR1 and resultant metabolites on a Caco-2 model. Samples were qualitatively analyzed using
HPLC-MS/MS and quantitatively determined using HPLC-UV. When incubated with pooled human
intestinal bacteria anaerobically, NGR1 showed biphasic elimination: an insignificant decrease in the
first 8 h followed by a rapid elimination during 8-48 h. Four metabolites, three unambiguously
identified as ginsenosides Rg1, F1 and 20(S)-protopanaxatriol formed via stepwise deglycosylation,
and one tentatively assigned as a dehydrogenated protopanaxatriol with transformation occurring at
the tetracyclic triterpenoid skeleton, were produced sequentially. Rg1 and F1 were formed tran-
siently at low apparent velocities, while 20(S)-protopanaxatriol was the major metabolite with a
formation rate close to the rate of NGR1 elimination and a low elimination rate. NGR1 remained
intact in human liver S9 or microsomes over 1 h. Transport study of NGR1 and its metabolites
revealed an ascending permeability order with stepwise deglycosylation. Taken together, the results
revealed a determinant role of intestinal bacteria in the overall disposition and potential bioactivity of
NGR1 in human.
KEYWORDS: Radix notoginseng; notoginsenoside R1; permeability; metabolism; intestinal bacteria;
Caco-2
INTRODUCTION
Radix notoginseng (Chinese name Tienchi or Sanqi) is the
dried root of Panax notoginseng (Burk.) F. H. Chen, an herb
noted for its blood circulation promotion, blood stasis removal
and pain alleviation effects, and has been widely utilized for the
prevention and treatment of microcirculatory disturbance in
China and other Asian countries for many years (1). Radix
notoginseng is classified as a functional food in China (2), and
in the United States and European countries various Radix
notoginseng based herbal products are available on the market
as dietary supplements.
Similar to P. ginseng C. A. Meyer (Asian ginseng) and
P. quinquefolius L. (American ginseng), P. notoginseng belongs
to the Ginseng genus and contains protopanaxadiol (PPD)- and
protopanaxatriol (PPT)-type saponins as its main compo-
nents (3-5). Notoginsenoside R1 (NGR1, Figure 1) is one of
the PPT-type saponins in P. notoginseng. Compared to other
Ginseng herbs, P. notoginseng contains NGR1 in much more
substantial amount (5). Thus, NGR1 is widely adopted as the
characteristic marker of P. notoginseng when distinguishing this
herb from other Ginseng herbs and evaluating quality of Radix
notoginseng or related products (3-6). Extensive pharmacologi-
cal evaluation of NGR1 has been conducted on both in vivo and in
Figure 1. Chemical structures of notoginsenoside R1 and its main meta-
bolites formed by human intestinal bacteria; glc, glucose; xyl, xylose.
vitro models. NGR1 has exhibited various health promotion
activities, including neuroprotective (7), hepatic protective (8),
cardiovascular protective (9, 10), and immune stimulation
effects (11). Therefore, NGR1 was considered as one of the main
bioactive ingredients contributing to beneficial effects of Radix
notoginseng and documented in China Pharmacopoeia as the
chemical marker for quality control of this herb (12).
In contrast to the well-defined beneficial actions, so far, the
information on the absorption, distribution, metabolism and
*To whom correspondence should be addressed. E-mail: ruyan@
umac.mo. Tel: 853-83974876. Fax:853-28841358.
pubs.acs.org/JAFC
Published on Web 04/21/2010
© 2010 American Chemical Society

Article
J. Agric. Food Chem., Vol. 58, No. 9, 2010 5771
excretion (ADME) properties and consequently the in vivo fate of
NGR1 is surprisingly limited. Recent studies with Radix noto-
ginseng extract revealed a very low oral bioavailability (<10%)
of NGR1 in the rat (13-15). However, the oral pharmacokinetic
property of individual NGR1 in human remains to be addressed.
On the basis of the existing reports on other PPD- and PPT-type
saponins, a poor permeability in the gut and an extensive
hydrolysis in the intestinal tract of NGR1 are anticipated after
oral dosing, thus leading to a low oral bioavailability. On the other
hand, it is widely accepted that, when taken orally, ginsenoside
metabolites formed by intestinal bacteria, instead of the parent
saponins, reach the systemic circulation easier and hence exert
beneficial effects (16, 17). However, so far, there is little evidence
supporting this speculation in the case of NGR1. Moreover, the
role of the liver, the principal site of biotransformation of most
drugs, in the in vivo process of NGR1 is still unclear.
Therefore, the present study was designed to (1) characterize
NGR1 hydrolysis by human intestinal bacteria, qualitatively and
quantitatively; (2) determine and compare the permeability
properties of NGR1 and metabolites generated by intestinal
bacteria on a Caco-2 monolayer model; and (3) examine meta-
bolic stability of NGR1 in human liver with S9- and microsome-
based incubation systems.
incubations or reactions without microflora solution or NGR1 served as
controls. Each reaction was performed 3 times. Reactions were stopped by
addition of 2 mL of ice-cold 1-butanol followed by immediate centrifuga-
tion at 5000g for 30 min to remove the bacteria. After adding 100 μL of
ginsenoside Rb1 (10 mM) as internal standard, the sample was twice
extracted with 10 mL of water saturated 1-butanol. The organic fractions
was combined and evaporated under N2 at 37 °C. The residue were then
reconstituted with 200 μL of methanol, filtered through a 0.45 μm
membrane filter before being subjected to HPLC analysis.
Calibration Curves of NGR1, Ginsenosides Rg1, F1 and PPT in
Human Intestinal Bacterial Incubation System. Stock solutions of
NGR1, Rg1, F1 and PPT were prepared and diluted to appropriate
concentrations with DMSO for construction of calibration curves. Each
calibration curve contained 6 different concentrations and was performed
in triplicate. An aliquot (100 μL) of stock solutions of each compound at
different concentrations was mixed with 4.9 mL of BHI medium contain-
ing 100 μL of human microflora solution prepared as described above. The
resultant samples were processed immediately as described above. The
calibration curves were constructed by plotting the peak area ratio of the
spiked analyte to internal standard as a function of the concentration
of each analyte. The limits of detection (LOD) and quantification (LOQ)
under the present chromatographic conditions were determined at a
signal-to-noise ratio (S/N) of above 3 and 10, respectively.
Metabolic Stability of NGR1 in Human Hepatic Subcellular
Fractions. The metabolic stability of NGR1 in human liver subcellular
fractions was determined in a total of 200 μL of reaction solution
containing NGR1 (final concentration 0.2 mM), human liver S9 or
microsomes (1 mg/mL), NADPH-regenerating system (4 mM of MgCl2,
1 mM NADPþ, 1 mM glucose-6-phosphate and 1 U/mL glucose-6-
phosphate dehydrogenase) and 100 mM potassium phosphate buffer
(pH 7.4). Reactions were conducted at 37 °C for 60 min. The incubation
systems with denatured liver S9 or microsome or without NADPH-
regenerating system served as controls. All the experiments were per-
formed in triplicate. Reactions were terminated by adding 200 μL of
methanol and vortexing to mix thoroughly. After centrifugation at 15000g
for 5 min, the resultant supernatant was filtered through a 0.45 μm mem-
brane filter and an aliquot (70 μL) applied to an HPLC-DAD system or
an HPLC-MS/MS system.
Culture of Caco-2 Cells. Caco-2 cells obtained from the American
Type Culture Collection at passage 30-40 were cultured in DMEM
supplemented with 10% fetal bovine serum and 1% nonessential amino
acids, at 37 °C in an atmosphere of 5% CO2 and 90% relative humidity.
Cells were subcultured at 80-90% confluence by trypsinization with
0.05% trypsin-EDTA. In transport studies, Caco-2 cells (at passages
30-50) were seeded on 12-well plates at a density of 1 ꢀ 105 cells/well and
cultured for 21 days before starting transport study. For each experiment,
the integrity of the monolayer was monitored by measuring the trans-
epithelial electrical resistance(TEER) (19) with an epithelial voltohmmeter
(World Precision Instruments, Inc., FL) and the permeability of the
paracellular marker lucifer yellow. Only Caco-2 monolayers with TEER
above 300 Ω cm2 before and after the transport study and a leakage rate of
lucifer yellow less than 1% per hour were utilized in the transport study.
The efflux transporter P-glycoprotein (P-gp) functionality in Caco-2
monolayers was validated using the P-gp probe substrate rhodamine
123. Rhodamine 123 (5 μM) exhibited substantial directional preference
with an efflux ratio of 39.4, suggesting the normal presence of P-gp
transporter in the Caco-2 cell monolayers used.
MATERIALS AND METHODS
Materials. Notoginsenoside R1 (purity >95%) was supplied by
Kuiqing Trading Co. Ltd. (Tianjin, China). Ginsenoside F1 (purity
>98%) and 20(S)-protopanaxatriol (PPT) (purity >98%) were pur-
chased from National Institute for the Control of Pharmaceutical and
Biological Products (Beijing, China). Ginsenoside Rg1 and Rb1 were
isolated from Radix notoginseng in our institute with purities >95% by
HPLC analysis. BBL brain heart infusion (BHI) medium, GasPak EZ
Anaerobe Container System with Indicator and GasPak EZ Large Incu-
bation Container were purchased from Becton Dickinson (Franklin
Lakes, NJ). L-Cystine was from Research Organics, Inc. (Cleveland, OH).
Hanks’ balanced salt solution (HBSS), rhodamine 123, collagen type I
from rat tail, sodium pyruvate, dimethyl sulfoxide (DMSO), hemin bovine
and vitamin K1 were supplied by Sigma-Aldrich (St. Louis, MO).
Methanol, 1-butanol and acetonitrile were HPLC-grade from Merck
(Darmstadt, Germany). Deionized water was purified by a Milli-Q
purification system (Millipore; Bedford, MA). Dulbecco’s modified
Eagle’s medium (DMEM), fetal bovine serum (FBS), 0.25% trypsin-
EDTA, penicillin-streptomycin solution and nonessential amino acids
were purchased from GibcoBRL Life & Technologies (Grand Island,
NY). Transwell plates (12-well, 0.4 μm pore size, 1.12 cm2, polycarbonate
filter) were purchased from Corning Costar Co (Cambridge, MA). Caco-2
cells were obtained from the American Type Culture Collection (Rockville,
MD). Pooled human liver S9 and microsomes were supplied by Sigma.
Preparationof HumanIntestinal Bacteria. The culture medium was
modified from the method reported by Chang and Nair (18). In brief, 100
mL of autoclaved BHI medium (3.7 g/100 mL) was supplemented with
0.05 mg of vitamin K1, 0.5 mg of hemin bovine and 50 mg of L-cystine.
Fresh fecal samples from healthy Chinese volunteers (20-32 years) were
provided by Kiang Wu hospital (Macau, China) according to a protocol
approved by the Medical Department, Kiang Wu Hospital and the
Internal Ethical Committee of the Institute of Chinese Medical Sciences,
University of Macau. Five fecal samples (2 g of each) were pooled together
and mixed well with 30 mL of culture medium. The resultant fecal
suspension was centrifuged at 200g for 5 min and supernatant decanted
and centrifuged at 5000g for 30 min. The resultant precipitate was
resuspended with 10 mL of BHI medium to produce intestinal microflora
solution.
Metabolism of NGR1 by Human Intestinal Bacteria. The bio-
transformation of NGR1 by human intestinal bacteria was determinedin a
5 mL incubation system containing 250 μL of intestinal microflora
solution and 100 μL of NGR1 stock solution in DMSO (NGR1 final
concentration 0.2 mM) in BHI medium. The incubation system was
anaerobically incubated at 37 °C in a GasPak EZ Anaerobe Pouch System
for 0, 4, 8, 12, 15, 18, 21, 24, 36, 48, and 72 h, respectively. Zero-minute
Transport of NGR1, Ginsenosides Rg1, F1 and PPT across Caco-
2 Cell Monolayer. Prior to the transport study, cytotoxicity of NGR1,
ginsenoside Rg1, F1 and PPT toward Caco-2 cells was determined using
MTT assays. Non-cytotoxic concentrations of NGR1 (1 mM), Rg1
(1 mM), F1 (200 μM) and PPT (50 μM) were chosen for transport study.
Transport study was carried out in HBSS buffer. After 21 days of
culture, the prepared Caco-2 monolayers were rinsed twice with HBSS and
preincubated in HBSS at 37 °C for 30 min. In the absorption transport
study, 0.5 mL of HBSS solutions containing NGR1, Rg1, F1 or PPT was
loaded at the apical (A) side (donor chamber), and 1.5 mL of blank HBSS
was placed at the basolateral (B) side (receiver chamber). In the secretion
transport study, 1.5 mL of the HBSS containing test compound was added
at the B side (donor chamber) and 0.5 mL of blank HBSS was placed at the
A side (receiver chamber). Aliquots of 0.1 mL were taken from receiver

5772 J. Agric. Food Chem., Vol. 58, No. 9, 2010
Ruan et al.
Figure 2. Typical HPLC chromatograms of(A) 12 h control reaction without NGR1, (B) 12 h reaction with NGR1, (C) 48 h control reaction without NGR1, (D)
48 h reaction with NGR1. *: impurity detected in the incubation system with time prolonged.
chambers at different time intervals (0, 30, 60, 90, 120 min) for HPLC
analysis. After each sampling, 0.1 mL of HBSS was added to the receiver
chamber to maintain a constant volume. At the last time point, 0.1 mL of
sample was also withdrawn from the donor chamber for HPLC analysis to
calculate the recovery. All the experiments were performed three times in
duplicate.
The apparent permeability coefficients (Papp) of NGR1, Rg1, F1 and
PPT from apical side to basolateral side (PappAfB) or from basolateral
side to apical side (PappBfA) in bidirectional transport study on Caco-2
cell model were calculated using the following equation as reported
previously (20):
50 mM ammonium acetate in water (A) and acetonitrile (B). The gradient
was the same as described above in the section HPLC-DAD Analysis.
ESI-MS analyses were operated in both positive and negative ionization
modes. Negative ion mode provided higher sensitivity, better reproduci-
bility and less interference from the incubation system under the present
analytical conditions and, thus, was chosen for sample analysis in the
present study. The detailed conditions for negative ionization mode were
as follows: drying gas N2 8 L/min, temperature 325 °C, pressure of nebulizer
30 psi, capillary voltage -3500 V, scan range 100-1400 m/z. ESI-MS/MS
conditions were as follows: negative ion mode, separation width 4,
fragment amplification 1.0, scan range 100-1400 m/z.
Data Analysis. All data were expressed as mean ( standard deviation
(SD). Difference between permeability of NGR1 and its metabolites was
assessed using unpaired Student’s t-test. A p < 0.05 value was deemed
significant for all tests.
Papp ¼ ðdC=dt ꢀ VÞ=ðC0 ꢀ AÞ
where dC/dt is the rate of the test compound appearing in the receiver
chamber, V is the volume of the solution in the receiver chamber, C0 is the
initial concentration of the test compound added in the donor chamber,
and A is the cell monolayer surface area.
RESULTS
Efflux ratio (PappBfA/PappAfB) >2 was adopted when determining
whether efflux transporter(s) was involved in transport of the test
compound.
Metabolism of Notoginsenoside R1 by Human Intestinal Bacteria.
When NGR1 was incubated with pooled human intestinal
bacteria, the peak of NGR1 decreased with time and several
additional peaks were observed within 72 h incubation (Figure 2).
When compared with controls (a mixture of NGR1 with intest-
inal bacteria without incubation, incubations without NGR1,
and NGR1 in BHI medium alone without bacteria), four of these
unknown peaks were identified as metabolites of NGR1 with
retention times of 38.1 min (M1), 46.5 min (M2), 52.6 min (M3)
and 54.3 min (M4) (Figures 2B and 2D).
The structures of M1, M2 and M3 were unambiguously
identified as ginsenoside Rg1, F1 and 20(S)-protopanaxatriol
(PPT) by comparison of their retention times, UV spectra and
mass spectra (MS1 and MS2) with those of the authentic com-
pounds. As shown in Figure 3, the mass spectrum of M1 showed
the pseudo molecular ion [M - H]- at m/z 799, which was 132
mass units less than that of NGR1 (m/z 931) corresponding to the
loss of one xylose, and the acetoxy adduct ion [M þ AcO]- at m/z
HPLC-DAD Analysis. Quantitative analysis was completed on an
Agilent series 1200 HPLC apparatus (Agilent Technologies, Santa Clara,
CA) equipped with a vacuum degasser, a binary pump, an autosampler
and a diode array detector. An Alltech Alltima C18 column (250 mm ꢀ 4.6
mm, 5 μm) was used. The column temperature was maintained at 25 °C.
The flow rate was 1.5 mL/min. Injection volume was 5 μL of samples from
intestinal bacterial metabolic studies and 70 μL of other samples. NGR1,
ginsenosides Rg1, F1 and PPT were monitored at 203 nm, and their UV
absorption was recorded over 210-400 nm. The mobile phase consisted of
water (A) and acetonitrile (B), and a gradient elution was adopted as
follows: 0-30 min, 18-19% B; 30-40 min, 19-31% B; 40-60 min,
31-100% B.
HPLC-MS/MS Analysis. Structural identification was performed
on an LC/MSD ion trap system (Agilent Technologies, Palo Alto, CA)
consisting of an HP 1100 series binary pump HPLC system and an ion-
trap mass spectrometer with an electrospray ionization interface, connected
to an Agilent ChemStation software. The mobile phase consisted of

Article
J. Agric. Food Chem., Vol. 58, No. 9, 2010 5773
Figure 3. MS1 and MS2 data of NGR1 and its metabolites formed by human intestinal bacteria. M1: ginsenoside Rg1. M2: F1. M3: PPT. M4: DH-PPT.
859. Its MS/MS data showed ions at m/z 637 ([M - H - Glc]-)
and m/z 475 ([M - H - 2Glc]-), the same as those observed with
standard ginsenoside Rg1. In addition, the retention time and the
UV spectrum of M1 were also identical to those of ginsenoside
Rg1. Thus, M1 was identified as ginsenoside Rg1. The character-
istic ions at m/z 637 ([M - H]-) and m/z 697 ([M þ AcO]-) in the
mass spectrum of M2 indicated that this metabolite has a mole-
cular weight of 638 Da, which corresponded to loss of one glucose
from ginsenoside Rg1. Its UV and MS/MS spectra were identical
to those of F1 and Rh1 while the retention time was the same as
that of F1 but different from that of Rh1. Therefore M2 was
assigned as F1. The mass spectrum of M3 showed its [M - H]-
ion at m/z 475, [M þ HCOO]- ion at m/z 521, [M þ AcO]- ion at
m/z 535 and [2M - H]- ion at m/z 951, indicating a molecular
weight of 476 Da corresponding to further loss of one glucose
from F1. Its MS/MS spectrum also exhibited a characteristic
ion at m/z 391 ([M - H - C6H12]-) resulting from the neutral loss
of the aliphatic side chain at C20 position. Moreover, the
retention time and UV spectrum of M3 were the same as those
of 20(S)-protopanaxatriol. Thus, M3 was identified as 20(S)-
protopanaxatriol (PPT).
The identity of M4 was only tentatively assigned due to lack of
the standard compound. Mass spectrum of M4 exhibited char-
acteristic ions at m/z 473 ([M - H]-), m/z 519 ([M þ HCOO]-)
and m/z 533 ([M þ AcO]-), all 2 mass units less than those of
PPT, indicating a dehydrogenated product of PPT. Furthermore,
the appearance of M4 was later than PPT. Therefore, M4 was
tentatively identified as a dehydrogenated product of PPT.
Similar to PPT, the characteristic ion at m/z 389 corresponding
to the loss of the aliphatic side chain ([M - H - C6H12]-) was

5774 J. Agric. Food Chem., Vol. 58, No. 9, 2010
Ruan et al.
Table 1. Calibration Curves for NGR1 and Its Metabolites
y = ax þ b
Table 2. Apparent Permeability Coefficients (Papp) of NGR1 and Its
Metabolites Determined on the Caco-2 Modela
PappAfB
(ꢀ10-6 cm/s)
PappBfA
(ꢀ10-6 cm/s)
PappBfA/
PappAfB
retention
time
(min)
concn
range
compd
slope
(a)
intercept
(b)
detection
limit (μM)
compd
r2
(μM)
NGR1
Rg1
F1
0.088 ( 0.010
0.236*** ( 0.017
0.866## ( 0.220
18.170ξξξ ( 0.942
0.093 ( 0.011
0.244*** ( 0.017
4.523### ( 2.06
18.720ξξξ ( 2.417
1.05
1.03
5.22
1.03
NGR1
Rg1
F1
35.6
38.1
46.5
52.6
12.227
14.884
18.925
18.657
0.161
0.010
0.091
0.009
0.9999 2.2-280.0
0.9998 2.2-280.0
0.9996 1.4-87.5
1.0000 2.5-320.0
0.41
0.48
0.39
0.40
PPT
a Data are presented as means ( SD (n = 3). ***p < 0.001, ##p < 0.01 and
###p < 0.001, ξξξp < 0.001 as compared to the corresponding values for NGR1, Rg1
and F1, respectively.
PPT
were observed in the amounts of NGR1 left and PPT generated at
36 h due to a variation in NGR1 hydrolyzing rates over 8-48 h.
When the amounts of ginsenosides Rg1, F1 and PPT generated
were calculated and compared based on 1:1 stoichiometric con-
version, the metabolites at their maximum levels accounted for
6.3% (Rg1), 1.1% (F1) and 71.0% (PPT) of initial NGR1,
respectively. Furthermore, the total recovery of the compound
per se and the metabolites ginsenosides Rg1, F1 and PPT formed
before the advent of DH-PPT reached >95%. All these data
supported that NGR1 was stepwisely deglycosylated mainly
through the following pathway: NGR1 f Rg1 f F1 f PPT.
Metabolism of NGR1 by Human Hepatic Subcellular Fractions.
When incubated with human liver S9 fraction or microsomes in
the absence or presence of NADPH-regenerating system for 1 h,
NGR1 remained intact and the recovery of NGR1 reached 98 (
1.5% (S9) and 97 ( 2.8% (microsome), respectively. Further-
more, HPLC-MS/MS analysis revealed no phase I transforma-
tion of NGR1.
Transport of NGR1 and Its Main Metabolites across Caco-2
Monolayer. The bilateral Papp values for NGR1 and its meta-
bolites across Caco-2 cell monolayers are summarized in Table 2.
With stepwise removal of glycosyl groups, the Papp values
(PappAfB value in the case of F1) of the metabolites increased.
NGR1 exhibited a very low Papp value (ꢀ10-8 cm/s), indicatinga
very poor oral absorption (<1%) in human body. The Papp
values (ꢀ10-7 cm/s) of ginsenosides Rg1 and F1 (A to B side)
were 1 order of magnitude higher than that of NGR1, while PPT
showed a Papp (ꢀ10-5 cm/s) that was 2 orders of magnitude
higher than that of Rg1 and F1 (A to B side), indicating an
increasing permeability. The efflux ratios (PappBfA/PappAfB) of
NGR1, Rg1, and PPT were within the range of 1.0-1.5, suggest-
ing that there was no significant difference between the perme-
ability in the A to B and that in the B to A directions and NGR1,
ginsenosides Rg1, and PPT transport across intestinal epithelial
cells mainly through passive diffusion, whereas the transport of
F1 from B to A direction was significantly higher than that
obtained from A to B direction and the efflux ratio was 5.22,
indicating that F1 might be the substrate of efflux transporter(s).
Figure 4. Time courses of NGR1 metabolism and metabolite generation
by human intestinal bacteria. A: Peak area ratio-time plot. B: Concen-
tration-time plot.
observed, indicating that the dehydrogenation of PPT occurred
on the tetracyclic triterpenoid skeleton.
Time Course of NGR1 Metabolism by Human Intestinal Bacteria.
As shown in Figure 2, NGR1 and its metabolites were eluted with
baseline separation under the present chromatographic condi-
tion. The calibration curves (Table 1) for NGR1, ginsenoside
Rg1, F1 and PPT provided good linearity (r2 > 0.998) over the
concentration ranges tested. The detection limits were 0.4 μM
(NGR1, F1 and PPT) and 0.5 μM (Rg1) (Table 1).
The time courses of NGR1 metabolism and metabolite gene-
ration by human intestinal bacteria are shown in Figure 4. During
the first 8 h, NGR1 decreased very slowly (average velocity:
0.18 μmol/h) with >97% remained at 8 h, whereas it was
eliminated rapidly (average velocity: 4.17 μmol/h) afterward
and disappeared from the reaction system 48 h after incubation.
Rg1 and F1 reached their maximum at 8 and 15 h respectively and
then were eliminated rapidly, whereas PPT was not detected until
8 h, reached its maximum at 48 h (average velocity over 8-48 h:
3.40 μmol/h) and decreased slowly afterward with a slow increase
of the dehydrogenated metabolite M4 (Figure 4). Great variations
DISCUSSION
In the past decade, people have attributed the beneficial effects
of many ginsenosides to their metabolites formed in the gastro-
intestinal tract when taken orally. Although degradation path-
ways of many ginsenosides have been well examined in vivo and in
vitro using animals (15-17, 21-23), information in human is
sparely available.
As the characteristic constituent and the chemical marker of
Radix notoginseng, NGR1 exhibited diverse beneficial effects.
However, its ADME properties and hence its biologically active
form(s) are still unclear. In the present study, NGR1 metabolism
by human intestinal bacteria and human liver subcellular frac-
tions were investigated, for the first time, to predict its disposition
in humans.

Article
Four metabolites appeared stepwisely when NGR1 was in-
J. Agric. Food Chem., Vol. 58, No. 9, 2010 5775
form supports its high resistance to hepatic metabolism in the
rat (15). In the present study, there was no evidence supporting
significant biotransformation of NGR1 in human liver micro-
somes. These results imply that hepatic metabolism, if any, plays
an insignificant role in NGR1 disposition.
cubated with different pooled human intestinal bacteria, three of
which were unambiguously identified as ginsenoside Rg1, F1 and
20(S)-protopanaxatriol by comparison of their retention times,
UV spectra and MS spectra with those of standards. The initial
colonic deglycosylation of NGR1 occurred preferentially at C6 to
yield Rg1. The resultant Rg1 could further remove one molecule
of glucose either at C-6 to generate F1 or at C-20 to have Rh1. In
previous reports on Rg1 hydrolysis by rat intestinal bacteria, both
Rh1 and F1 formations were speculated, although no definite
evidence was provided to support Rh1 formation. Other studies
support that acidic hydrolysis of Rg1 in the stomach generated
Rh1 while F1 was formed from Rg1 by intestinal bacteria (21,22).
In the present study, existence of Rh1 in the human intestinal
bacterial incubation system was ruled out using Rh1 standard.
Another metabolite M4, which was tentatively identified as a
dehydrogenated PPT (DH-PPT) based on MS analyses, was
detected 18 h after incubation. To the best of our knowledge,
this is the first report of dehydrogenation of PPT by human
intestinal bacteria. The dehydrogenation position was on the
tetracyclic triterpenoid skeleton as evidenced by the characteristic
ion of the core structure of M4 two mass units less than that of
PPT, thus M4 might be formed via carbonylation at C3, C6 or
C12. However, in a recent report from Yang et al., a dehydroge-
nated PPT was detected in rat urine when ginsenoside Re was
administered and the authors speculated a dehydrogenation at
the aliphatic side chain of PPT on basis of similar MS/MS
data (23). The absolute structure of DH-PPT, its production in
vivo and biological activities are worthy of investigation.
The appearance and elimination of Rg1 and F1 were rapid and
their quantities detected in the incubation system were relatively
low (Figure 4). Furthermore, PPT formation rate was around
80% of NGR1 elimination rate (3.40 μmol/h vs 4.17 μmol/h) over
the period of 8-48 h, indicating that, once formed, Rg1 was
further transformed to F1 and then F1 to PPT very rapidly. PPT
was the most abundant metabolite after 12 h and reached >70%
of initial NGR1 at 48 h. PPT is the end metabolite of PPT-type
ginsenosides via stepwise deglycosylation. PPT exhibited anti-
angiogenic, radiation-protective and antioxidation effects (22-26),
and over 80% of PPT entered the circulation (27), supporting its
important role in vivo.
The transport of the main metabolites of NGR1 across Caco-2
cell monolayer was compared with the parent compound in
parallel. Our data showed a low permeability of NGR1 across
Caco-2, suggesting a poor absorption (<1%) in human after
taken orally. This result was in good agreement with the low oral
bioavailability (0.2-0.6%) of NGR1 in rats reported re-
cently (15), indicating a determinant role of NGR1 permeability
in its oralbioavailability. Metabolites formed by human intestinal
bacteria exhibited higher permeability than NGR1 in the follow-
ing increasing order: PPT > F1 > Rg1 > NGR1 (Table 2).
These findings suggest that the metabolites formed by in-
testinal bacteria may have better absorption and thus might
be the main in vivo forms when NGR1 is taken orally. Since
other factors including solubility, biliary excretion, acidic
hydrolysis in gastric juice may also affect the oral bioavail-
ability of PPT- and PPD-type ginsenosides, in vivo study is
warranted to determine the bioavailability of NGR1 and its
main drug-related forms after oral dosing.
The present study revealed that NGR1 has poor permeability,
no hepatic biotransformation, yet extensive hydrolysis by intest-
inal bacteria in gut lumen via NGR1 f Rg1 f F1 f PPT f DH-
PPT using human-derived in vitro models. The resultant deglyco-
sylated metabolites bear better permeability. These findings, if
true in vivo, indicate the implications of the metabolites formed by
deglycosylation in human when NGR1 is taken orally, thus
calling for future investigations on their in vivo fates in the human
body.
ABBREVIATIONS USED
NGR1, notoginsenoside R1; PPD, protopanaxadiol; PPT,
protopanaxatriol; DH-PPT, dehydrogynated protopanaxatriol.
ACKNOWLEDGMENT
The authors thank Jacky Lei (the Chief of Department of
Clinical Laboratory, Kiang Wu Hospital) and his colleagues for
their support in feces sample collection from healthy volunteers.
It is very interesting to note that there were at least two phases
of NGR1 elimination by human intestinal bacteria. In phase 1
(within the first 8 h of incubation), NGR1 was slowly deglyco-
sylated at an average velocity of 0.18 μmol/h. While in phase 2
(during 8-48 h), NGR1 disappeared from the system rapidly at
an average velocity (4.17 μmol/h) > 20 times of phase 1. The
apparent biphasic elimination was also observed with another
PPT-type ginsenoside Rg1 while absent with the PPD-type
ginsenoside Rb1, which showed an immediate rapid elimination
in human intestinal bacteria (our group unpublished data). This
might attribute to different intestinal microflora that are involved
in biotransformation of PPD- and PPT-type ginsenosides (28).
The mechanism governing the biphasic deglycosylation of NGR1
by human intestinal bacteria warrants further investigation.
Recent reports reveal that, in addition to the deglycosylation in
gastrointestinal tract, phase I metabolism also plays a role, but to
different extents, in in vivo fates of some ginsenosides (15,29). Lai
et al. (29) identified two mono-oxygenated metabolites of ginse-
noside Rh1 in vivo after intragastrical administration and oxida-
tion caused about 75% Rh1 loss within 30 min in rat hepatic S9
and microsomes. The monooxidized metabolite of Rg1 detected
in rat bile and urine only accounted for 0.2% and 0.1% of its oral
dose, respectively (15). In the case of NGR1, the observation that
NGR1 was mainly eliminated via biliary excretion in its intact
LITERATURE CITED
(1) Jia, L.; Zhao, Y.; Liang, X. J. Current evaluation of the millennium
phytomedicine- ginseng (II): Collected chemical entities, modern
pharmacology, and clinical applications emanated from traditional
Chinese medicine. Curr. Med. Chem. 2009, 16, 2924–2942.
(2) Yang, Y. X. Scientific substantiation of functional food health
claims in China. J. Nutr. 2008, 138, 1199S–1205S.
(3) Wang, C. Z.; Ni, M.; Sun, S.; Li, X. L.; He, H.; Mehendale, S. R.;
Yuan, C. S. Detection of adulteration of notoginseng root extract
with other Panax species by quantitative HPLC coupled with PCA.
J. Agric. Food Chem. 2009, 57, 2363–2367.
(4) Dong, T. T. X.; Cui, X. M.; Song, Z. H.; Zhao, K. J.; Ji, Z. N.; Lo,
C. K.; Tsim, K. W. K. Chemical assessment of roots of Panax
notoginseng in China: regional and seasonal variations in its active
constituents. J. Agric. Food Chem. 2003, 51, 4617–4623.
(5) Wan, J. B.; Li, S. P.; Chen, J. M.; Wang, Y. T. Chemical character-
istics of three medicinal plants of the Panax genus determined by
HPLC-ELSD. J. Sep. Sci. 2007, 30, 825–832.
(6) Qian, Z. M.; Wan, J. B.; Zhang, Q. W.; Li, S. P. Simultaneous
determination of nucleobases, nucleosides and saponins in Panax
notoginseng using multiple columns high performance liquid chromato-
graphy. J. Pharm. Biomed. Anal. 2008, 48, 1361–1367.
(7) Gu, B.; Nakamichi, N.; Zhang, W. S.; Nakamura, Y.; Kambe, Y.;
Fukumori, R.; Takuma, K.; Yamada, K.; Takarada, T.; Taniura, H.;

5776 J. Agric. Food Chem., Vol. 58, No. 9, 2010
Ruan et al.
Yoneda, Y. Possible protection by notoginsenoside R1 against
glutamate neurotoxicity mediated by N-methyl-D-aspartate recep-
tors composed of an NR1/NR2B subunit assembly. J. Neurosci. Res.
2009, 87, 2145–2156.
(19) Shah, P.; Jogani, V.; Bagchi, T.; Misra, A. Role of Caco-2 cell
monolayers in prediction of intestinal drug absorption. Biotechnol.
Prog. 2006, 22, 186–198.
(20) Artursson, P.; Karlsson, J. Correlation between oral drug absorption
in humans and apparent drug permeability coefficients in human
intestinal epithelial (Caco-2) cells. Biochem. Biophys. Res. Commun.
1991, 175, 880–885.
(21) Hasegawa, H.; Sung, J. H.; Matsumiya, S.; Uchiyama, M. Main
ginseng saponin metabolites formed by intestinal bacteria. Planta
Med. 1996, 62, 453–457.
(8) Chen, W. X.; Wang, F.; Liu, Y. Y.; Zeng, Q. J.; Sun, K.; Xue, X.; Li,
X.; Yang, J. Y.; An, L. H.; Hu, B. H.; Yang, J. H.; Wang, C. S.; Li,
Z. X.; Liu, L. Y.; Li; Zheng, J.; Liao, F. L.; Han, D.; Fan, J. Y.; Han,
J. Y. Effect of notoginsenoside R1 on hepatic microcirculation
disturbance induced by gut ischemia and reperfusion. World J.
Gastroenterol. 2008, 14, 29–37.
(9) Zhang, W.; Wojta, J.; Binder, B. R. Effect of notoginsenoside R1 on
the synthesis of tissue-type plasminogen activator and plasminogen
activator inhibitor-1 in cultured human umbilical vein endothelial
cells. Arterioscler. Thromb. 1994, 14, 1040–1046.
(10) Zhang, H. S.; Wang, S. Q. Notoginsenoside R1 from Panax
notoginseng inhibits TNF-alpha-induced PAI-1 production in
human aortic smooth muscle cells. Vasc. Pharmacol. 2006, 44,
224–230.
(11) Zhang, H. S.; Wang, S. Q. Notoginsenoside R1 inhibits TNF-alpha-
induced fibronectin production in smooth muscle cells via the ROS/
ERK pathway. Free Radical Biol. Med. 2006, 40, 1664–1674.
(12) The State Pharmacopoeia Commission of P.R. China. Radix et
rhizome notoginseng. In Pharmacopoeia of the People’s Republic
of China; Chemical Industry Press: Beijing, China, 2010; Vol. I,
pp 10-11.
(13) Li, X.; Wang, G.; Sun, J.; Hao, H.; Xiong, Y.; Yan, B.; Zheng, Y.;
Sheng, L. Pharmacokinetic and absolute bioavailability study of
total panax notoginsenoside, a typical multiple constituent tradi-
tional chinese medicine (TCM) in rats. Biol. Pharm. Bull. 2007, 30,
847–851.
(14) Li, X.; Sun, J.; Wang, G.; Hao, H.; Liang, Y.; Zheng, Y.; Yan, B.;
Sheng, L. Simultaneous determination of panax notoginsenoside
R1, ginsenoside Rg1, Rd, Re and Rb1 in rat plasma by HPLC/
ESI/MS: platform for the pharmacokinetic evaluation of total
panax notoginsenoside, a typical kind of multiple constituent tradi-
tional Chinese medicine. Biomed. Chromatogr. 2007, 21, 735–
746.
(15) Liu, H.; Yang, J.; Du, F.; Gao, X.; Ma, X.; Huang, Y.; Xu, F.; Niu,
W.; Wang, F.; Mao, Y.; Sun, Y.; Lu, T.; Liu, C.; Zhang, B.; Li, C.
Absorption and disposition of ginsenosides after oral administration
of Panax notoginseng extract to rats. Drug Metab. Dispos. 2009, 37,
2290–2298.
(16) Odani, T.; Tanizawa, H.; Takino, Y. Studies on the absorption,
distribution, excretion and metabolism of ginseng saponins. IV.
Decomposition of ginsenoside-Rg1 and -Rb1 in the digestive tract
of rats. Chem. Pharm. Bull. (Tokyo) 1983, 31, 3691–3697.
(17) Qian, T.; Jiang, Z. H.; Cai, Z. W. High-performance liquid chromato-
graphy coupled with tandem mass spectrometry applied for meta-
bolic study of ginsenoside Rb1 on rat. Anal. Biochem. 2006, 352,
87–96.
(22) Han, B. H.; Park, M. H.; Han, Y. N.; Woo, L. K.; Sankawa, U.;
Yahara, S.; Tanaka, O. Degradation of ginseng saponins under mild
acidic conditions. Planta Med. 1982, 44, 146–149.
(23) Yang, L.; Xu, S.; Liu, C.; Su, Z. In vivo metabolism study of
ginsenoside Re in rat using high-performance liquid chromatogra-
phy coupled with tandem mass spectrometry. Anal. Bioanal. Chem.
2009, 395, 1441–1451.
(24) Liu, F. Y.; Wang, J. N.; Yu, S. D.; Wang, B.; Zhang, J. D. Effect of
panaxatriol on hematogenesis and granulocyte-macrophage colony
stimulating factor in radiation injured mice. Saudi Med. J. 2007, 28,
1791–1795.
(25) Kwok, H. H.; Ng, W. Y.; Yang, M. S.; Mak, N. K.; Wong, R. N.;
Yue, P. Y. The ginsenoside protopanaxatriol protects endothelial
cells from hydrogen peroxide-induced cell injury and cell death by
modulating intracellular redox status. Free Radical Biol. Med. 2010,
48, 437–445.
(26) Usami, Y.; Liu, Y. N.; Lin, A. S.; Shibano, M.; Akiyama, T.;
Itokawa, H.; Morris-Natschke, S. L.; Kenneth Bastow, K.; Kasai,
R.; Lee, K. H. Antitumor agents. 261. 20(S)-Protopanaxadiol and
20(S)-protopanaxatriol as antiangiogenic agents and total assign-
ment of 1H NMR spectra. J. Nat. Prod. 2008, 71, 478–481.
(27) Cui, J. F.; Bjorkhem, I.; Eneroth, P. Gas chromatographic-mass
spectrometric determination of 20-(S)-protopanaxadiol and 20(S)-
protopanaxatriol for study on human urinary exc retion of ginseno-
sides after ingestion of ginseng preparations. J. Chromatogr., B:
Biomed. Sci. Appl. 1997, 689, 349–355.
(28) Hasegawa, H.; Sung, J. H.; Benno, Y. Role of human intestinal
Prevotella oris in hydrolyzing ginseng saponins. Planta Med. 1997,
63, 436–440.
(29) Lai, L.; Hao, H.; Liu, Y.; Zheng, C.; Wang, Q.; Wang, G.; Chen, X.
Characterization of pharmacokinetic profiles and metabolic path-
ways of 20(S)-ginsenoside Rh1 in vivo and in vitro. Planta Med. 2009,
75, 797–802.
Received for review February 11, 2010. Revised manuscript received
April 12, 2010. Accepted April 12, 2010. The present work was
supported by the Science and Technology Development Fund (Ref.
No.: 049/2005/A-R1), Macau SAR, the National Basic Research
Program of China (973 program, Grant No.: 2009CB522707) and the
Research Committee of the University of Macau (Ref. No.: RG086/07-
08S/WYT/ICMS).
(18) Chang, Y. C.; Nair, M. G. Metabolism of daidzein and genistein by
intestinal bacteria. J. Nat. Prod. 1995, 58, 1892–1896.